Even legislators with a keen interest in protecting the U.S. corporate tax base from profit shifting should probably resist the urge to jump into the middle of Microsoft’s high-profile transfer pricing dispute with the IRS.
Microsoft’s recent disclosure that the IRS believes the company paid $28.9 billion less in corporate income taxes than it owed from 2004 through 2013 has understandably drawn considerable interest, both from within and without the tax practitioner community.
Reasons for the dispute’s notoriety include the unprecedented magnitude of the alleged underpayments and the pivotal role played by cost-sharing arrangements (CSAs), the most contentious structure in U.S. section 482 litigation.
The dispute has also drawn outsize attention for the unusually antagonistic nature of the IRS’s examination of Microsoft, which was chronicled by ProPublica in a 2020 exposé, and the debate it sparked about the IRS’s use of private law firms in examinations.
Sen. Elizabeth Warren, D-Mass., is evidently among those who have taken note. In an October 17 letter to Microsoft Chair and CEO Satya Nadella, which gave Nadella all of 10 days to respond, Warren pressed the company’s top executive for details on the tax planning practices at issue and other information not disclosed or addressed in Microsoft’s October 11 Form 8-K or in a related blog post.
The statements in Warren’s letter and the questions put to Microsoft are primarily based on ProPublica’s 2020 piece and its recent follow-up article, which generally portray the IRS as outmatched and enfeebled in its ill-fated clash with one of the world’s most powerful corporate giants.
Some of Warren’s questions probe the company’s decision not to disclose the total interest or penalties associated with its alleged tax underpayments, which are valid issues to raise considering companies’ documented tendency to downplay the extent of their exposure in U.S. transfer pricing disputes.
However, the IRS’s basis for calculating penalties and interest isn’t inside information. Given the amount of the allegedly underpaid tax and the circumstances of the case, it’s probably safe to assume that the agency seeks to impose a 40% gross valuation misstatement penalty under section 6662(h) on substantially all of the alleged underpayment.
Assuming the IRS does so, Microsoft’s alleged tax underpayments and penalties would total about $40.5 billion. The interest that Microsoft may owe would vary depending on the distribution of the proposed adjustments across a decade’s worth of tax years, but it would almost certainly be material.
The vintage of the 10-year period, which began two decades ago, and the enhanced interest rates applicable to large corporate underpayments under section 6621(c), suggest that interest accrued through September 2023 could have more than doubled the amount Microsoft allegedly owes.
Even a grand total in proposed deficiencies, penalties, and interest of over $90 billion is within the range of plausible estimates.
However, most of the letter’s other statements, questions, and innuendos are premature or off the mark. This includes Warren’s apparent assumption that the $28.9 billion figure reported by Microsoft is somehow the last word in a case that hasn’t even come before the IRS Independent Office of Appeals, much less been heard by the Tax Court. As Warren observes in her letter:
“These are deeply disturbing disclosures, identifying what appears to be an egregious example of corporate tax evasion and misbehavior. They reveal that, for nearly a decade, Microsoft may have systematically underpaid its taxes by billions of dollars — rewarding shareholders and executives, while depriving the federal government of revenue needed to pay for health care, environmental protection, national defense, and more — and leaving middle class taxpayers to foot the bill.” [Internal citation omitted.]
The government’s revenue needs obviously don’t dictate the legality of Microsoft’s tax arrangements, and the amount the company owes won’t be finalized for years to come. That’s not to say that there aren’t good reasons to believe that Microsoft owed far more in U.S. corporate tax than it paid for the years at issue; there most certainly are.
But just because Microsoft’s calculation of its tax liability was likely wrong doesn’t mean the IRS’s computations were necessarily 100% right. How can we say either way before the IRS Appeals process has run its course?
For this reason alone, Microsoft can’t be reasonably faulted for challenging the IRS’s assessment through the appeals process.
However, Warren evidently sees the matter differently:
“You owe Congress and the public an explanation for your actions, particularly given reports that Microsoft intends to appeal and continue fighting the results of the audit.” [Emphasis added.]
Even those who consistently sympathize with the IRS in cost-sharing disputes should recognize that the amount stated in a notice of proposed adjustment isn’t etched in clay tablets. Maybe the IRS overshot by 10% and Microsoft really underpaid its taxes by $26.3 billion instead of $28.9 billion.
If the IRS seeks to impose a 40% section 6662(h) penalty on the entire shortfall, correcting the agency’s slightly overzealous calculation would reduce the amount Microsoft owes in taxes and penalties by about $3.7 billion before interest.
And a conservative estimate of the interest assessed on that amount could more than double the total reduction of what the company ultimately has to pay, reducing the bill by $7.4 billion.
Not even Microsoft and its investors should turn up their noses at that kind of savings. The company would be justified in challenging the amount of the IRS’s proposed adjustments, even if it were only to establish that the agency was slightly off in percentage terms.
Suggesting that Microsoft owes the public an explanation for even trying to vet the IRS’s proposed adjustments through the appeals process is unreasonable.
It’s also necessary to acknowledge the reality that courts have often sided with taxpayers in transfer pricing cases the IRS should have won. There are still those who strongly disagree that reg. section 1.482-4(b)’s definition of “intangible” invalidated the IRS’s discounted cash flow valuation in Amazon.com Inc. v. Commissioner, 148 T.C. No. 8 (2017), aff’d, 934 F.3d 976 (9th Cir. 2019), but the law is ultimately what courts say it is.
Facebook is now testing a revised version of Amazon’s argument in the Tax Court (Facebook Inc. v. Commissioner, No. 21959-16), and even a 2009 regulatory overhaul painstakingly devised to make such arguments impossible may not guarantee the IRS victory.
In Microsoft’s case, there’s yet another wild card: Early signs suggest that the IRS may be making an argument that hasn’t yet been tested in court. It’s entirely possible that an occasionally error-prone Tax Court will reject a well-founded argument that it hasn’t encountered before.
If there’s a non-trivial likelihood that a judge would take Microsoft’s side in the dispute, and there probably is, then the company’s decision to contest proposed adjustments of such magnitudes isn’t really cause for a public apology.
Even if the IRS’s calculation of underpaid tax were absolute, final, and correct, Microsoft could still reasonably challenge any assessment of accuracy-related penalties under section 6662. Imposing a 40% section 6662(h) gross valuation misstatement penalty on Microsoft’s alleged underpayment would amount to $11.6 billion, and the amount owed could easily double after accounting for interest.
Considering the inherent subjectivity of the standards set by reg. section 1.6662-6(d), a penalty challenge by Microsoft is fair game. Under reg. section 1.6662(d)(2)(ii), penalties don’t apply to any net section 482 adjustments attributable to a controlled transaction for which “the taxpayer selects and applies a specified method in a reasonable manner.” Whether a taxpayer applied a specified method in a reasonable manner must be determined based on all the facts and circumstances, including the list of considerations identified in reg. section 1.6662(d)(2)(ii)(A)(1) through (7).
The factors identified in the regulations include the taxpayer’s experience and knowledge, the reasonableness of its search for reliable data, and its compliance with the regulatory standards applicable to the selected method.
Microsoft used what was then a specified method for pricing a buy-in payment, the residual profit-split method, to determine the buy-ins for its CSAs at issue in the dispute.
Although few would suggest that Microsoft is inexperienced or uninformed in tax planning, the exhaustiveness of its search for data and the reliability of its residual profit-split method analysis are subjective questions with no indisputable answer. And the analysis could be further complicated if the IRS’s legal basis for the adjustments is novel.
Another factor relevant for penalty assessment purposes is the taxpayer’s reliance on qualified advisers. Microsoft obviously relied on professionals qualified to provide tax advice in setting up its CSAs, and Warren’s letter reflects a misguided fixation on the individual firm that provided the advice for one of them: KPMG. (Full disclosure: I worked for KPMG in 2013 and 2014.)
This is on display in the letter’s questions about when Microsoft began taking tax advice from KPMG, whether the company followed all of KPMG’s recommendations, and whether Microsoft still receives tax advice from the firm.
What KPMG pitched and what Microsoft acted on are now largely matters of public record, so it’s hard to see the value in pushing Microsoft to confirm what the IRS and everyone else already knows.
It’s also strange to press Microsoft to recall KPMG tax advice it ultimately rejected and to disclose any ongoing tax advisory relationship the company may have with the firm. The only apparent point in asking these questions, it would seem, is to insinuate that KPMG is uniquely tainted among firms that provide international tax planning advisory services.
If Microsoft had switched from KPMG to another Big Four firm, would it have made any difference? KPMG may have been the firm to conceive of Microsoft’s most aggressive CSA, which was admittedly the rightful subject of an intensive IRS examination.
But there’s little reason to believe that KPMG, almost 20 years after the arrangement at issue was put in place, is any more complicit in aggressive tax planning than its peers.
If anything, KPMG may be slightly underrepresented as advisers to the arrangements challenged by the IRS in major U.S. transfer pricing disputes. A different Big Four firm calculated the CSA buy-in payment challenged by the IRS in Amazon. And yet another Big Four firm valued the platform contribution transaction in the ongoing Facebook case, priced the intercompany royalty disputed in Medtronic Inc. v. Commissioner, T.C. Memo. 2022-84, and advised on what the IRS considers a sham transaction in Perrigo Co. v. United States, 294 F. Supp. 3d 740.
Taxpayers interested in aggressive intellectual property offshoring advice have many vendors to choose from, and there’s no valid reason to construe a relationship with KPMG as evidence of impropriety.
Warren’s letter, referring to the 2020 ProPublica exposé, also accuses Microsoft of attempting “to derail the IRS and its ability to identify corporate tax evaders.” Microsoft clearly tried to thwart the IRS’s examination, and it did so in many different ways.
But the letter’s excerpt from the 2020 ProPublica article, which refers to unspecified changes in law and restrictions on the IRS’s examination tools, along with Warren’s questions for the company’s CEO, suggest a focus on Microsoft’s calls for congressional intervention:
“Is Microsoft continuing to lobby Congress or the Administration — either directly or through membership associations or other third parties — for policies that would allow the company to evade paying the full amount of unpaid taxes identified in the IRS audit?” [Emphasis added.]
Oddly, Warren fails to identify the practice specifically targeted by Microsoft and taxpayer advocacy groups: The IRS’s engagement of a private law firm to assist it in collecting facts, reviewing taxpayer data, and questioning witnesses during an examination.
The IRS’s decision to hire litigation firm Quinn Emanuel Urquhart & Sullivan LLP to support its examination of Microsoft’s CSAs sparked a major controversy that, at least until recently, overshadowed the underlying transfer pricing dispute.
Some of us may regard Microsoft’s concerns as overblown, but the IRS’s ability to outsource examination functions is a valid policy question on which reasonable minds could differ. Even U.S. District Court Judge Ricardo S. Martinez, who rejected Microsoft’s claim that the IRS’s contract with Quinn Emanuel improperly delegated an inherently governmental function, expressed reservations with the arrangement.
In his November 2015 order in United States v. Microsoft Corp., No. 2:15-cv-00102 (W.D. Wash.), which granted the government’s motion to enforce the IRS summonses at issue, Martinez noted that he was “troubled by Quinn Emanuel’s level of involvement” and correctly predicted that it would draw a response from Congress.
The concerns voiced by Microsoft and shared by many others should be at least vaguely familiar to Warren, who touted her influence on the version of the Taxpayer First Act that ultimately cleared the Senate without objection in a June 2019 voice vote. Warren’s principal interest in the bill was to ensure that the IRS wasn’t barred from creating its own free online filing system, a commendable consumer protection initiative she has championed for years.
But the Taxpayer First Act, as amended to address Warren’s concerns, also added a new code section that severely restricts the IRS’s power to outsource examination functions.
The addition of section 7602(f), a measure that would be hard to overlook in a bill that only runs 37 pages, codified the very prohibition that Microsoft had wished for and that taxpayer lobbying groups demanded. It’s a bit misleading for Warren’s letter to portray Microsoft’s lobbying efforts as sprawling and nefarious when they targeted a specific practice and culminated in a single bill that Warren apparently supported.
It’s also odd for Warren to neglect the fate of Microsoft’s other attempts to thwart the examination. After Martinez’s 2015 order granting the government’s petition to enforce the IRS’s summonses, Microsoft still tried to withhold 174 documents sought by the IRS.
The company claimed that the documents were protected by some combination of the work product doctrine, attorney-client privilege, or the federally authorized tax practitioner privilege. However, this obstruction campaign ended in resounding defeat: Martinez’s January 2020 order in Microsoft directed the company to hand over the remaining documents, and the company’s subsequent Freedom of Information Act litigation efforts were rejected in 2023.
In other words, from a tax enforcement perspective, the system worked. Aside from prompting a nearly unanimous congressional decision to prohibit the IRS from engaging private law firms to assist in examinations, Microsoft’s great derailment campaign came to naught.
Despite its best efforts, Microsoft was ultimately ordered by a U.S. district court judge to give the IRS the information the company had fought so hard to withhold. And based on the massive figure cited in Microsoft’s Form 8-K, the IRS made good use of what it got. Now that the summonses have been enforced and the notices sent, the agency’s need for belated reinforcement by individual senators is unclear.
What goes for the summons enforcement litigation probably goes for the underlying transfer pricing dispute as well. Warren’s letter relies on an excerpt from the 2020 ProPublica article observing that the IRS had “become notably less bold” since 2019, but the narrative that Microsoft overpowered a cowed IRS seems a bit out of date as of October 2023.
Claiming an unprecedented $28.9 billion in underpaid tax, plus interest and penalties, doesn’t really seem like something a helpless tax administration would do.
Direct congressional intervention, either through individual questioning or a formal public hearing, in a matter already subject to enforcement proceedings by the responsible executive agency, may offer enticing grandstanding opportunities.
But the risks of legislative meddling generally outweigh the potential rewards unless the case reveals some systemic agency enforcement breakdown, official misconduct, or problem that requires new legislation to fix. Warren obviously isn’t suggesting the IRS acted unethically, and it would be bizarre to construe Microsoft’s disclosure as a sign that the agency has become powerless to carry out its mission using existing legislative tools.
Perhaps future developments in the dispute will justify legislators’ involvement. If, for example, the IRS Office of Appeals were to withdraw the examiners’ transfer pricing adjustments entirely, that might be a worthy topic of congressional interest. A disconnect of that magnitude would suggest that the examination team, the Appeals officer, or both were billions of dollars off the mark.
In the meantime, it could be useful for legislators to take a closer look at the obstacles the IRS generally faces in transfer pricing enforcement or to confirm that the agency is appropriately and effectively using every statutory and regulatory tool at its disposal.
However, jumping into Microsoft’s transfer pricing dispute with the IRS at this stage in the process is premature. For the moment, the best legislative response to a series of historically large IRS transfer pricing adjustments is probably to take note and let the process play out.